Download presentation
Presentation is loading. Please wait.
Published byValentine Stewart Modified over 8 years ago
1
Using Test Delegations from the Root Prior to Full Allocation and Delegation DNS-OARC Fall workshop, October 2013 Andrew Sullivan Principal Architect
2
Pg. 2 Lessons from Internet Protocols What is this all about? If we want to know whether some candidate TLD is in use on the Internet, we have to measure that.
3
Pg. 3 Lessons from Internet Protocols The name collision report http://www.icann.org/en/about/staff/secu rity/ssr/name-collision-02aug13-en.pdf
4
Pg. 4 Lessons from Internet Protocols The draft http://tools.ietf.org/html/draft-kolkman- root-test-delegation-00 (updates soon)
5
Pg. 5 Lessons from Internet Protocols Disclaimer Draft written with Olaf Kolkman, Geoff Huston, George Michaelson, and Warren Kumari They didn’t vet this presentation & when I get it wrong, please just blame me
6
Pg. 6 Lessons from Internet Protocols Overview Basic approach is to stimulate queries for the candidate string At the same time, stimulate queries for some other (regular) name Compare the behaviour
7
Pg. 7 Lessons from Internet Protocols What we’ll need Some special names A special name server An advert network The candidate string in the root (!)
8
Pg. 8 Lessons from Internet Protocols Special labels TestName. CandidateTLD. ExistingTLD.
9
Pg. 9 Lessons from Internet Protocols The TestName A constant This string is used in every test. Random string Constant through test Should be “semantic nonsense” Must be LDH-valid Mustn’t exist already
10
Pg. 10 Lessons from Internet Protocols The ExistingTLD A control The ExistingTLD is used to hold names that should work fine on the Internet Needs to be “normal” Probably not com. Probably not an IDNA label Are there problems using a ccTLD?
11
Pg. 11 Lessons from Internet Protocols The CandidateTLD The subject The CandidateTLD’s use is what we’re trying to test Assume the low-risk category from the report is right Do this for the other cases Needs to be delegated to instrumented name server
12
Pg. 12 Lessons from Internet Protocols The An identifier The uniqueid ties the different queries together Each test round gets a uniqueid that identifies the same query source Syntactically valid LDH label
13
Pg. 13 Lessons from Internet Protocols QNAMEs we might see -a.TestName. -b.TestName.CandidateTLD. -c.TestName.ExistingTLD.
14
Pg. 14 Lessons from Internet Protocols Cases One Local use of CandidateTLD but no search list - b.TestName.CandidateTLD. doesn’t exit the network -a.TestName. and - c.TestName.ExistingTLD. will be observed
15
Pg. 15 Lessons from Internet Protocols Cases Two Local use of CandidateTLD and CandidateTLD in search list - b.TestName.CandidateTLD. doesn’t exit the network -a.TestName. gets search listed, doesn’t exit the network - c.TestName.ExistingTLD. will be observed
16
Pg. 16 Lessons from Internet Protocols Cases Three No local use of CandidateTLD, including in search lists All names observed
17
Pg. 17 Lessons from Internet Protocols Cases Four No local use of CandidateTLD, including in search lists - b.TestName.CandidateTLD. and - c.TestName.ExistingTLD both observed -a.TestName. gets search list: - a.TestName.CandidateTLD
18
Pg. 18 Lessons from Internet Protocols Open questions Should we test for http fetches? – Requires actual resolution of names Is even a test delegation acceptable? How much use is ok? – Probably a policy question that’s not our problem?
19
Using Test Delegations from the Root Prior to Full Allocation and Delegation Thank you
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.